Cell Surface Accumulation of a Truncated Transmembrane Prion Protein in Gerstmann-Straussler-Scheinker Disease P102L*

A familial prion disorder with a proline to leucine substitution at residue 102 of the prion protein (PrP 102L ) is typically associated with protease-resistant PrP fragments (PrP Sc ) in the brain parenchyma that are infectious to recipient animals. When modeled in transgenic mice, a fatal neurodegenerative disease develops, but, unlike the human counterpart, PrP Sc is lacking and transmission to recipient animals is questionable. Alternate mice expressing a single copy of PrP 102L (mouse PrP 101L ) do not develop spontaneous disease, but show dramatic susceptibility to PrP Sc isolates from different species. To understand these discrepant results, we studied the biogenesis of human PrP 102L in a cell model. Here, we report that cells expressing PrP 102L show decreased expression of the normal 18-kDa fragment on the plasma membrane. Instead, a 20-kDa fragment, probably derived from transmembrane PrP ( Ctm PrP), accumulates on the cell surface. Because the 20-kDa fragment includes an amyloidogenic region of PrP that is disrupted in the 18-kDa form, increased surface expression of 20-kDa fragment may enhance the susceptibility of these cells to PrP Sc infection by providing an optimal substrate, or by amplifying the neurotoxic signal of PrP Sc . Thus, altered susceptibility of PrP 101L mice to exogenous PrP Sc may be mediated by the 20-kDa Ctm PrP 6 ). A small amount of the 20-kDa fragment is also detected ( lane 6 ). PrP 102L lysate, on the other hand, reveals the three PrP glycoforms, but the 18-kDa fragment and its 22-kDa glyco- sylated form are barely detectable ( lane 7 ). Deglycosylation shows the 27-kDa full-length form as above, but in contrast to PrP C , the 18-kDa fragment is 5-fold less, and the 20-kDa is 4-fold more in PrP 102L lysates ( lane 8 ). A stronger 20-kDa band is observed in 3F4 blots because of higher affinity of 3F4 as compared with 8H4 antibody

Familial prion disorders of humans segregate with mutations in the prion protein gene (PRNP). Based on the clinicopathological presentation, these disorders are categorized as Gerstmann-Straussler-Scheinker disease (GSS), 1 Creutzfeldt-Jakob disease, and fatal familial insomnia (reviewed in Refs. [1][2][3][4]. A proline to leucine substitution at codon 102 of PRNP is one of the most frequent mutations associated with GSS, and also one of the earliest prion disorders to be ascribed a genetic etiology (5,10). Most GSS cases present with cerebellar ataxia and dementia, and a prominent pathological feature is the presence of multicentric amyloid plaques composed of protease resistant PrP fragments of 8, 15, and 21-30 kDa. Although the 21-kDa fragment has also been observed in Creutzfeldt-Jakob disease, the 8-kDa fragment appears specific to GSS (6 -8). The 21-30-kDa fragments are more prominent in GSS102L with a spongiform change, and low molecular mass fragments of 8 or 7-15 kDa are predominant in cases with multicentric amyloid plaques (7)(8)(9)(10). These fragments are generated in vivo and are believed to contribute to the neuropathology observed in these disorders.
Modeling of GSS102L in transgenic mice expressing the corresponding PrP mutation (MoPrP P101L) has produced unusual and provocative observations. Animals expressing high levels of the transgene showed spontaneous neurodegeneration at 85-300 days of age without the presence of detectable PKresistant PrP Sc (11). The observed neurodegeneration in these mice is believed to result from adverse effects of the mutant PrP rather than overexpression of the transgene (12). Intracerebral inoculation of tissue from diseased animals transmitted the disease to healthy P101L mice expressing low levels of the transgene, but not to mice expressing normal PrP (11,12). Subsequently, mice expressing a single copy of the P101L allele were generated by homologous recombination to avoid artifacts because of transgene overexpression. These mice also lacked PrP Sc deposits, but surprisingly, although none of the animals developed spontaneous disease, their susceptibility to exogenous PrP Sc infection was altered dramatically (13,14). Thus, it appears that PrP 102L alters certain cellular characteristics or functions of neuronal cells that influence the replication or toxicity of exogenously introduced PrP Sc . The reasons for these discrepant observations are presently unclear.
We investigated the biogenesis of PrP 102L in transfected human neuroblastoma cells in an attempt to model the events that might occur in vivo, and uncover the biochemical pathways of neurotoxicity in GSS102L. In this report, we show that the metabolism of PrP 102L is altered, resulting in accumulation of a 20-kDa fragment of PrP on the cell surface, with a concomitant decrease in the expression of 18-kDa fragment, a product of normal recycling of PrP from the plasma membrane (15). The 20-kDa fragment is likely derived from Ctm PrP, a transmembrane form of PrP that has been implicated as a mediator of neurodegeneration in certain inherited and infectious prion disorders (21,22). Our results suggest that the change in ratio of 18:20-kDa fragments on PrP 102L cells may increase their vulnerability to PrP Sc toxicity by unconventional pathways, thus accounting for the complex biological effects of PrP102L in vivo.

EXPERIMENTAL PROCEDURES
Cell Lines, Reagents, and Antibodies-The human neuroblastoma cell line M17 was obtained from Dr. J. Beidler (Memorial Sloan-Kettering Cancer Center, New York, NY). Opti-MEM, fetal calf serum, penicillin/streptomycin, methionine, and cysteine-free Dulbecco's modified Eagle's medium (DMEM), and LipofectAMINE were from Invitrogen; hygromycin B and MG132 were from Calbiochem; Tran 35 S-label was from ICN (Costa Mesa, CA); protein A-agarose and N-glycosidase F (PNGase-F) were from Roche Molecular Biochemicals; phosphatidylinositol-specific phospholipase C (PI-PLC) was from GLYKO (Novato, CA). All other chemicals were purchased from Sigma. Transfected M17 cells expressing wild type (PrP C ) or mutant (PrP 102L ) prion protein were generated as described in a previous report (17,18). All cultures were maintained at 37°C in Opti-MEM supplemented with 5% fetal calf serum and 1% penicillin-streptomycin, in a humidified atmosphere containing 5% CO 2 . Cultures of transfected cells were supplemented with 500 g/ml hygromycin to maintain the episomal plasmid carrying the PrP cDNA. Experiments were performed on bulk-selected cells at different times after transfection. For all experiments, cells were replated overnight and used at 90 -95% confluence. The following antibodies were used in this study: anti-PrP monoclonal antibodies 3F4 (R. Kascsak, New York State Institute for Basic Research in Developmental Disabilities, Staten Island, NY) and 8H4 (M. S. Sy, Department of Pathology, Case Western Reserve University, Cleveland, OH), and anti-PrP immune serum 2301 (generated in our facility). The antibody 3F4 is specific for residues 109 and 112 of PrP, 8H4 binds to an epitope between residues 145-180 of PrP, and 2301 antiserum was raised to C-terminal 220 -230 residues of PrP.
Western Blotting-In a typical experiment, 9 ϫ 10 6 cells were used for each condition. Equal amount of total protein was used from cells expressing either PrP C or PrP 102L . Cells were rinsed with PBS and lysed in a buffer containing 0.5% Nonidet P-40, 0.5% deoxycholate, and 10 mM EDTA in Tris-buffered saline (TBS, 20 mM Tris, 150 mM NaCl, pH 7.4), containing 10 g/ml each of leupeptin, antipain, pepstatin, and 1 mM phenylmethylsulfonyl fluoride (PMSF). Cell debris was cleared by centrifugation at 290 ϫ g, and protein in the supernatant was precipitated with five volumes of cold methanol at Ϫ20°C for 2 h. Total cellular proteins were fractionated by SDS-PAGE and electrophoretically transferred to Immobilon-P (Millipore) for 2.5 h at 70 V at 4°C. Membranes containing transferred proteins were blocked in TBS containing 10% nonfat dry milk and 0.1% Tween 20 for 1 h at 37°C and probed with anti-PrP antibodies 3F4 (1:40,000), 8H4 (1:1000), or 2301 (1:1000) dissolved in antibody dilution buffer (TBS, 1% normal goat serum, 0.05% bovine serum albumin) essentially as described (18). Immunoreactive bands were reacted with anti-mouse antibody conjugated to horseradish peroxidase (1:4000), and visualized on an autoradiographic film by ECL (Amersham Biosciences). To detect PI-PLCcleaved PrP in the culture medium, cells were treated with PI-PLC as described below, and the released PrP was methanol-precipitated and detected as described above.
Metabolic Labeling and Immunoprecipitation-In a typical experiment, 9 ϫ 10 6 cells expressing PrP C or PrP 102L were used. Immunoprecipitation was performed essentially as described (17,19). Cells were starved in methionine-cysteine-free DMEM containing 5% dialyzed serum for 1 h, in the presence or absence of the appropriate inhibitor, and labeled with 0.166 mCi/ml Tran 35 S-label in the same medium for 30 min or 2 h. For overnight labeling, the labeling medium was mixed with normal medium containing serum in a ratio of 3:1. Labeled cells were washed with PBS and lysed as described above for Western blotting. Cell debris was cleared by centrifugation at 290 ϫ g, and clarified lysates were subjected to immunoprecipitation with the appropriate antibodies in the presence of 1% bovine serum albumin and 0.1% N-lauryl sarcosine. Protein-antibody complexes bound to protein Aagarose were washed four times with 0.5 ml of wash buffer (150 mM NaCl, 10 mM Tris-HCl, pH 7.8, 0.1% N-lauryl sarcosine, and 0.1 mM PMSF), the bound protein was eluted by boiling in sample buffer (Tris-HCl, pH 6.8, 3% SDS, 10% glycerol, 5% ␤-mercaptoethanol) and analyzed by SDS-PAGE-fluorography. For pulse-chase experiments, cells were labeled with trans-[ 35 S]methionine for 30 min and chased in normal medium. At the indicated times, the cells were washed with PBS, lysed, and subjected to immunoprecipitation as above.
Labeling with [ 3 H]Ethanolamine-To identify the glycosyl phosphatidylinositol (GPI) anchor, cells were labeled with 250 Ci/ml [ 3 H]ethanolamine overnight in complete medium and processed for immunoprecipitation as above.
Labeling at 15°C-Two 10-cm dishes containing 9 ϫ 10 6 PrP C -or PrP 102L -expressing cells were washed with methionine-cysteine free DMEM, and preincubated in the same medium containing 5% dialyzed serum for 1 h at 37°C. Cellular proteins were metabolically labeled with 0.332 or 0.166 Ci/ml Tran 35 S-label at 15°C in a refrigerated incubator, or at 37°C in the absence or presence of 30 M MG132 for 2 h, and subjected to immunoprecipitation as above.
PI-PLC Digestion-Cells at steady state, or radiolabeled with trans-[ 35 S]methionine were washed with Opti-MEM without serum, and incubated with 0.066 -0.099 units/ml PI-PLC in fresh medium for 1 h at 37°C. The medium was collected and centrifuged at 4°C for 10 min at 290 ϫ g to pellet any cells. PI-PLC-released proteins in the supernatant were precipitated with five volumes of cold methanol at Ϫ20°C for 2 h, fractionated by SDS-PAGE, and immunoblotted or immunoprecipitated as described above with the appropriate antibody to detect PrP.
Enzymatic Deglycosylation-For deglycosylation, unlabeled, or radiolabeled immunoprecipitated proteins were reprecipitated with five volumes of cold methanol and resuspended in denaturing buffer (0.5% SDS, 1% ␤-mercaptoethanol). Samples were boiled for 10 min and deglycosylated with PNGase-F (1000 units in 1% Nonidet P-40, 25 mM sodium phosphate, pH 7.5) for 1-3 h at 37°C. Proteins were reprecipitated with five volumes of cold methanol at Ϫ20°C for 2 h, dissolved in sample buffer, and resolved by SDS-PAGE immunoblotting or fluorography.
Cell Homogenization and Treatment with Proteinase K-PrP C -and PrP 102L -expressing cells were washed and homogenized on ice in a buffer containing 10 mM HEPES, pH 7.9, 1.5 mM MgCl 2 , 10 mM KCl, and 0.5 mM dithiothreitol with 20 strokes of a Kontes all-glass Dounce homogenizer. The homogenate was checked microscopically for cell breakage and centrifuged to pellet nuclei. The resulting supernatant was centrifuged at 20,000 ϫ g to pellet membrane vesicles. These were resuspended in 0.5 ml of transport buffer (25 mM HEPES, pH 7.4, 115 mM KOAc, 2.5 mM MgCl 2 , 10 mM KCl, 2.5 mM CaCl 2 , and 1 mM dithiothreitol), and treated with 20 g/ml proteinase K on ice for 30 min. After adding 5 mM PMSF to stop the reaction, membrane vesicles were re-pelleted, solubilized in lysis buffer, and immunoblotted with 3F4.
Assay of PrP Endocytosis-Cells expressing PrP C or PrP 102L were cultured on poly-D-lysine-coated glass coverslips overnight. One set of PrP C and PrP 102L cells were incubated with anti-PrP antibody 3F4 (1:25) in complete medium for 30 min on ice, and two other sets of cells were incubated with the same antibody at the same dilution for 10 or 60 min each at 37°C in a humidified CO 2 incubator. At the end of each incubation period, cells were washed three times with PBS and fixed with 4% paraformaldehyde for 30 min at room temperature. Free aldehyde groups were quenched with 50 mM NH 4 Cl (in PBS), and nonspecific sites were blocked with PBS containing 10% goat serum and 0.2% bovine serum albumin, followed by 0.2% gelatin in PBS. Cells were then incubated with anti-mouse TRITC to detect the anti-PrP antibody bound to PrP molecules on the cell surface. Subsequently, the cells were incubated with 0.25 mg/ml unlabeled anti-mouse IgG for 40 min to block the remaining anti-PrP antibody sites on the cell surface, and permeabilized with 0.1% Triton X-100 for 4 min. The cells were rinsed in PBS, and the internalized anti-PrP antibody was reacted with FITCconjugated anti-mouse antibody to detect anti-PrP-PrP complexes that have been internalized from the cell surface. After a final wash in PBS, cells were mounted in gel mount (Biomeda Corp., Fostar City, CA) and observed using a laser scanning confocal microscope (Bio-Rad). 102L Cells-The steady state expression of PrP C and PrP 102L in transfected neuroblastoma cells was evaluated by immunoblotting cell-associated PrP with antibodies specific to PrP residues 109 and 112 (3F4), 145-180 (8H4), or 220 -230 (2301). Immunoreaction with 3F4 showed, as expected, three glycoforms of PrP C comprising the unglycosylated form of 27 kDa, intermediate forms of 29 -30 kDa, and highly glycosylated forms of 33-42 kDa (Fig. 1, lane 1). Deglycosylation with PNGase-F resulted in the appearance of a major species of 27 kDa, and a minor band migrating at 20 kDa. Lysates of PrP 102L showed similar glycoforms of PrP and the 20-kDa fragment, but in contrast to PrP C , all the PrP 102L glycoforms and the 20-kDa fragment migrated ϳ1 kDa faster on SDS-PAGE ( Fig. 1, lanes 1 and 2 versus lanes 3 and 4). More importantly, the 20-kDa fragment is 4-fold more in PrP 102L lysates as compared with PrP C (lane 2 versus lane 4).

Surface Expression of 18-and 20-kDa C-terminal Fragments of PrP Is Altered in PrP
Immunoblotting with 8H4 revealed, in addition to the three glycoforms of PrP C observed above, truncated forms migrating at 27-30, 22, and 18 kDa (Fig. 1, lane 5). These forms represent C-terminal fragments of the highly glycosylated (27-30 kDa), intermediate (22 kDa), and unglycosylated (18 kDa) forms that are a product of normal recycling of PrP from the cell surface (15). Deglycosylation with PNGase-F revealed a major form of 27 kDa representing full-length PrP, and the truncated fragment of 18 kDa. A small amount of the 20-kDa fragment was also detected (Fig. 1, lane 6). Similar processing of PrP 102L lysates revealed the three glycoforms of PrP, but surprisingly, the 18-kDa fragment and its glycosylated form of 22 kDa were barely detectable (Fig. 1, lane 7). Deglycosylation showed the 27-kDa full-length PrP as observed for PrP C . However, as compared with PrP C , the 18-kDa fragment was 5-fold less, and the 20-kDa fragment 4-fold more in PrP 102L lysates. Unlike the 20-kDa fragment, migration of 18-kDa fragment of PrP C and PrP 102L was similar ( Fig. 1, lanes 6 and 8). A stronger 20-kDa band was observed in 3F4 blots as a result of higher affinity of 3F4 as compared with 8H4 antibody. The apparent difference in pattern of PrP forms in the 3F4 and 8H4 blots is a result of the fact that, in addition to the full-length forms, 8H4 detected C-terminal PrP forms that result from cleavage at residue 111/112 of PrP during normal recycling from the plasma membrane. The C-terminal fragments of 18, 24, and 27-30 kDa that arise from this cleavage have been described in the literature, and do not immunoreact with 3F4 because of disruption of its epitope at residue 109 (15). The 20-kDa fragment is detected by both 3F4 and 8H4 antibodies, suggesting that it is a C-terminal fragment of PrP. The C-terminal antibody 2301 yielded similar results as obtained with 8H4 (data not shown).
The increased representation of 20-kDa fragment in PrP 102L lysates could be the result of increased production resulting from abnormal metabolism of PrP 102L or, conversely, decreased turnover of the 20-kDa fragment caused by misfolding or aggregation because it probably includes the PrP 102L mutation. The decreased representation of 18-kDa fragment, on the other hand, could be caused by reduced expression of full-length PrP 102L on the cell surface because of sequestration in an intracellular compartment or, conversely, normal surface expression but decreased endocytosis and/or recycling to the plasma membrane. These possibilities were investigated in the following experiments.
Over-representation of 20-kDa Fragment in PrP 102L Cells Is Not the Result of Increased Production-To evaluate whether accumulation of the 20-kDa fragment is the result of increased production as a result of aberrant metabolism of PrP 102L , PrP Cand PrP 102L -expressing cells were labeled with Tran 35 S-label for 2 h and subjected to immunoprecipitation with 3F4 or 8H4 antibodies. The three glycoforms of PrP and the 20-kDa fragment were detected with 3F4 in both PrP C and PrP 102L lysates, and as observed in Fig. 1, PrP 102L glycoforms and the 20-kDa fragment migrated ϳ1 kDa faster than PrP C forms ( Fig. 2A, lanes 1 and 2). The 20-kDa fragment was prominent in both samples, but in contrast to the results obtained in Fig. 1 at steady state, the 20-kDa fragment was not over-represented in PrP 102L lysates ( Fig. 2A, lanes 1 and 2). Similar results were obtained with 8H4 antibody ( Fig. 2A, lanes 3 and 4). A small amount of 18-kDa fragment could be detected in 8H4 immunoprecipitates, but there was no significant difference in the amount detected in PrP C and PrP 102L lysates ( Fig. 2A, lanes 3  and 4). Because the difference in the amount of 18-and 20-kDa fragments in PrP C and PrP 102L lysates was detected at steady state but not after a 2-h pulse, it probably arose as a result of a cumulative process that becomes apparent over time, and not because of an acute abnormality in the metabolism of PrP 102L .
To confirm the above results, the synthesis and turnover of PrP C and PrP 102L were compared in a pulse-chase paradigm. Cells were radiolabeled with Tran 35 S-label for 30 min and chased for 0, 2, and 4 h. At the end of each chase period, radiolabeled GPI-linked surface proteins were cleaved with PI-PLC at 37°C for 1 h, and the medium was collected. Subsequently, cell lysates and the PI-PLC-cleaved proteins were subjected to immunoprecipitation with 3F4 and fractionated by SDS-PAGE. Lysates treated with PI-PLC for 1 h soon after the pulse show the three glycoforms of PrP C and the 20-kDa fragment (Fig. 2B, lane 1). With increasing chase time, the amount of full-length PrP C forms and the 20-kDa fragment decreased in the lysate samples, and a corresponding increase was observed in the PI-PLC-cleaved samples (Fig. 2B, lanes 1-3 and  7-9). PrP 102L samples showed similar kinetics of synthesis and transport of full-length and 20-kDa forms to the plasma membrane (Fig. 2B, lanes 4 -6 and 10 -12), except for a significant decrease in full-length PrP 102L forms after 4 h of chase both in the lysate and PI-PLC-cleaved samples (Fig. 2B, lanes 6 and  12). In addition, a 14-kDa form was detected in the PrP 102L lysates soon after the pulse, which decreased gradually with chase (Fig. 2B, lanes 4 -6). Although the 14-kDa form of PrP 102L was not detected in the PI-PLC-cleaved samples, subsequent experiments with long term labeling showed that it was indeed secreted into the culture medium (see Fig. 4B).
The above results highlight important characteristics of the 20-kDa fragment: 1) it is generated soon after pulse and does not increase with chase, indicating that it is not a metabolic product of full-length PrP, 2) it is transported to the cell surface and is cleavable by PI-PLC, and 3) it does not accumulate intracellularly or on the cell surface after 4 h of chase.
Together, these results show that the 20-kDa fragment . Deglycosylation shows the 27-kDa full-length form as above, but in contrast to PrP C , the 18-kDa fragment is 5-fold less, and the 20-kDa is 4-fold more in PrP 102L lysates (lane 8). A stronger 20-kDa band is observed in 3F4 blots because of higher affinity of 3F4 as compared with 8H4 antibody arises from an alternate form of PrP, probably the transmembrane PrP ( Ctm PrP), and the accumulation of this fragment observed in PrP 102L lysates in Fig. 1 is not the result of increased production or sequestration in an intracellular compartment, but perhaps of reduced degradation.
If the 20-kDa fragment arises from Ctm PrP, it must be GPIlinked (20). To check this assumption, PrP C and PrP 102L cells labeled overnight with the anchor component [ 3 H]ethanolamine were lysed, immunoprecipitated with 3F4, and treated with PNGase-F to remove all glycans. The samples were fractionated on a long SDS-PAGE gel to accentuate the difference in migration of the 20-kDa fragment from PrP C and PrP 102L lysates. As expected, the 27-kDa full-length form and the 20-kDa fragment of PrP 102L migrated ϳ1 kDa faster on SDS-PAGE (Fig. 2C, lane 1 versus lane 2). More importantly, the 20-kDa fragment from both PrP C and PrP 102L lysates was labeled with [ 3 H]ethanolamine, confirming that it is linked with the GPI anchor (Fig. 2C, lanes 1 and 2). As observed in Fig. 1 above, the 20-kDa was 4-fold more in PrP 102L lysates as compared with PrP C .
The 20-kDa Fragment Is Generated from Transmembrane PrP in the Endoplasmic Reticulum-To evaluate whether the 20-kDa fragment arises from Ctm PrP, PrP C and PrP 102L cells were treated with the proteasomal inhibitor MG132 for 2 h, and subjected to immunoblotting with 3F4. There was a marked increase in the 20-kDa fragment after proteasomal inhibition, whereas the full-length PrP glycoforms were virtually unchanged (Fig. 3A, lane 1

versus lane 2 and lane 3 versus lane 4).
Because Ctm PrP is GPI-linked and degraded by the proteasomal pathway (20), an increase in a GPI-linked 20-kDa fragment (Fig. 2C) after proteasomal inhibition indicates its origin from Ctm PrP. A significant increase in a 26-kDa band was also noted after proteasomal inhibition, especially in PrP C lysates. The identity of this PrP form is currently under investigation.
To further confirm the origin of 20-kDa fragment from Ctm PrP, microsomes prepared from PrP C and PrP 102L cells were treated with 20 g/ml PK on ice for 30 min and subjected to immunoblotting with 3F4 or anti-calnexin antibodies. Because the N terminus of Ctm PrP faces the cytosol, it would be cleaved by PK treatment, releasing a C-terminal fragment of ϳ20 kDa in the ER, whereas the fully translocated PrP in the ER lumen would be protected from protease digestion and therefore show no change in migration or intensity (21,22). As shown in Fig. 3B, there was a small but significant increase in the 20-kDa fragment after protease digestion of PrP C and PrP 102L microsomes, whereas the full-length PrP glycoforms were virtually unchanged (Fig. 3B, lane 1 versus lane 2 and  lane 3 versus lane 4). Similar analysis of cell homogenates prepared from cells cultured with the proteasomal inhibitor MG132 for 2 h showed increased expression of the 26-and 20-kDa fragments in both PrP C and PrP 102L preparations (Fig.  3B, lanes 5-8). PrP 102L samples, in addition, showed a doublet in the region of 14 kDa when proteasomal function was inhibited (Fig. 3B, lane 7). PK treatment of microsomes prepared from these cells showed an increase in the 20-kDa fragment in both PrP C and PrP 102L samples, whereas the 26-kDa fragment and full-length PrP forms were unaffected (Fig. 3B, lane 5  versus lane 6 and lane 7 versus lane 8). Simultaneous cleavage of calnexin by PK to produce a faster migrating species lacking the cytosolic C-terminal domain confirms the efficacy of PK treatment and intactness of the microsomes in this experimental system (Fig. 3B, lanes 2, 4, 6, and 8). The above results strongly suggest that the 20-kDa fragment in our cell model is a product of Ctm PrP. The small increase in the amount of 20-kDa fragment after PK treatment is consistent with previous reports indicating that only ϳ2% of PrP is in the transmembrane orientation in cultured cells (23).
To evaluate whether up-regulation of Ctm PrP synthesis in PrP 102L cells accounts for the over-representation of 20-kDa fragment, PrP C and PrP 102L cells were radiolabeled in the  1 and 2), and, as observed in Fig. 1, PrP 102L glycoforms and the 20-kDa fragment migrate ϳ1 kDa faster than PrP C forms. The 20-kDa fragment is prominent in both samples, but is not over-represented in PrP 102L lysates (lanes 1 and 2). Similar results are obtained with 8H4 (lanes 3 and 4). B, PrP C and PrP 102L cells were labeled with Tran 35 S-label for 30 min and chased for 0, 2, and 4 h. At the end of each chase period, cells were subjected to PI-PLC treatment at 37°C for 1 h, and the medium was collected. The cell lysates and PI-PLC cleaved proteins were subjected to immunoprecipitation with 3F4 and fractionated by SDS-PAGE. Lysates of samples treated with PI-PLC soon after the pulse show the three glycoforms of PrP C and the 20-kDa fragment (lane 1). With increasing chase time, the amount of full-length PrP C forms and the 20-kDa fragment decrease in the lysate samples, and a corresponding increase is observed in the PI-PLC cleaved samples (lanes 1-3 and 7-9). PrP 102L samples show similar kinetics of synthesis and transport of full-length and 20-kDa forms to the plasma membrane (lanes 4 -6 and 10 -12). Note the decrease in full-length PrP 102L after 4 h of chase in the lysate and PI-PLC-cleaved samples (lanes 6 and 12). In addition a 14-kDa form is detected in the PrP 102L lysate soon after pulse, which decreases gradually with chase (lanes 4 -6). C, PrP C and PrP 102L cells were labeled overnight with the anchor component absence or presence of the proteasomal inhibitor MG132 for 2 h at 37°C, or at 15°C to block transport of proteins from the ER. Cell lysates were subjected to immunoprecipitation with 3F4 and analyzed by SDS-PAGE-fluorography. There was a small increase in all full-length PrP forms in the presence of MG132 both at 37°C and at 15°C (Fig. 3C, lanes 1-4 versus lanes 5-8), consistent with a recent report that ϳ10% of PrP C is degraded by the proteasomes (24). Cells labeled at 15°C, as expected, showed glycoforms with only high mannose core glycans. The highly glycosylated forms that are acquired in post-ER compartments were absent (Fig. 3C, lanes 2, 4, 6, and 8). The 20-kDa fragment was detected even at 15°C in both PrP C and PrP 102L lysates, confirming that it is generated in the ER (Fig.  3C, lanes 2 and 4). Surprisingly, an additional fragment of 14 kDa was detected at 15°C in PrP 102L lysates (Fig. 3C, lane 4). Both the 14-and 20-kDa fragments increased in amount in the presence of the proteasomal inhibitor MG132, indicating either increased production caused by a sparing of their precursor molecule or decreased turnover of these fragments by the proteasomes.
Quantitative estimation of the above results in terms of percentage increase of 20 kDa in comparison to full-length PrP forms shows increased accumulation of the 20-kDa fragment at 15°C, and in the presence of MG132 in both PrP C and PrP 102L lysates. The percentage of 20 kDa in PrP 102L lysates was less than PrP C under all four conditions examined, i.e. at 37°C or 15°C, and in the absence or presence of MG132. The 14-kDa fragment appears to be highly unstable, because it was detected only at 15°C or in the presence of MG132 and was significantly increased in PrP 102L cells. This fragment did not increase in amount after PK treatment of microsomes, arguing against its origin from the N-transmembrane form of PrP ( Ntm PrP) (21,22). The origin of this fragment is presently under investigation.
Together, the above results show that: 1) the 20-kDa fragment is generated in the ER, most likely from proteolytic cleavage of Ctm PrP at the ER membrane, 2) inhibition of Ctm PrP degradation by proteasomal inhibition results in an increase in the generation of 20-kDa fragment, 3) an additional N-terminal fragment of 14 kDa is detected in PrP 102L cells, the identity of which is presently unclear, and 4) increased representation of 20-kDa fragment in PrP 102L is not because of increased synthesis of Ctm PrP, but rather because of decreased turnover of this fragment, probably at the cell surface. In the following experiments, cell surface expression and turnover of the 20-kDa fragment were investigated to understand the cellular processes that lead to its accumulation in PrP 102L cells.
The 20-kDa Fragment Accumulates on the Surface of PrP 102L Cells-As noted in Fig. 1, not only is the 20-kDa over-represented in PrP 102L at steady state, but the 18-kDa fragment of PrP is significantly decreased. To study the metabolism of 18and 20-kDa fragments under conditions resembling steady state, PrP C -and PrP 102L -expressing cells were labeled with Tran 35 S-label overnight in a 3:1 ratio of methionine-cysteinefree and normal DMEM containing 5% dialyzed serum. After 18 h of radiolabeling, the labeling medium was collected to check for any secreted proteins, and cells were washed with PBS and incubated with PI-PLC for 1 h to cleave cell surface GPI-linked proteins. The cells were subsequently lysed, and the lysates, PI-PLC-cleaved proteins, and medium samples were subjected to immunoprecipitation with 3F4 or 8H4 and analyzed by SDS-PAGE-fluorography.
As expected, in the lysate and PI-PLC-cleaved PrP C samples, the three glycoforms representing the highly glycosylated, intermediate, and unglycosylated forms were detected. The 20-kDa fragment was mostly recovered in the PI-PLC-cleaved material rather than the lysate (Fig. 4A, lanes 1 and 3). PrP 102L lysates showed under-representation of the unglycosylated form as compared with PrP C , and all the PrP 102L forms migrated ϳ1 kDa faster than PrP C (Fig. 4A, lane 2). As observed in the PrP C samples, the 20-kDa fragment was mostly detected in the PI-PLC-cleaved material, and was over-represented in PrP 102L as compared with PrP C (Fig. 4A, lanes 3 and 4). The difference in the amount of 20-kDa fragment between PrP C and  8). C, PrP C or PrP 102L cells were radiolabeled in the absence or presence of the proteasomal inhibitor MG132 for 2 h at 37°C, or at 15°C to block transport of proteins from the ER. Cell lysates were subjected to immunoprecipitation with 3F4 and analyzed by SDS-PAGE fluorography. There is a small increase in all full-length PrP forms in the presence of MG132 both at 37°C and at 15°C (lanes 1-4 versus lanes [5][6][7][8]. Cells labeled at 15°C show glycoforms with only high mannose core glycans. The highly glycosylated forms that are acquired in the post-ER compartments are absent (lanes 2, 4, 6, and 8). The 20-kDa fragment is detected even at 15°C in both PrP C and PrP 102L lysates, confirming that it is generated in the ER (lanes 2 and 4). An additional fragment of 14 kDa is detected at 15°C in PrP 102L lysates (lane 4). D, quantitative estimation of the above results in terms of percentage increase of 20 kDa in comparison to full-length PrP forms shows increased accumulation of 20-kDa fragment at 15°C, and in the presence of MG132 in both PrP C and PrP 102L lysates.
PrP 102L samples became more prominent following deglycosylation of PI-PLC-cleaved samples, indicating that a significant proportion of this fragment on the cell surface is glycosylated, and migrates at 20 kDa when glycans are removed (Fig. 4A,  lanes 5 and 6). In addition, a 14-kDa fragment was detected in PI-PLC-cleaved proteins obtained from PrP 102L cells (Fig. 4A,  lane 4).
Immunoprecipitation with 8H4 showed similar PrP C and PrP 102L glycoforms as observed with 3F4. In addition, the 18-kDa C-terminal fragment of PrP that reacts with 8H4 but not with 3F4 was observed in PrP C lysates (Fig. 4A, lane 7), but the majority of this fragment was present in the PI-PLC-cleaved sample (Fig. 4A, lane 9). As noted in Fig. 1, the 18-kDa fragment was under-represented and the 20-kDa fragment was over-represented in PrP 102L lysates (Fig. 4A, lanes 8 and 10). Deglycosylation of PI-PLC cleaved samples showed 5-fold less 18-kDa and 4-fold more 20-kDa fragment in PrP 102L compared with PrP C (Fig. 4A, lanes 11 and 12). Immunoprecipitation of medium collected from PrP C and PrP 102L cells with 3F4 showed the presence of 14-kDa fragment in PrP 102L (Fig. 4B, lane 2), indicating that this fragment is secreted into the medium. A small amount of the full-length PrP C and PrP 102L forms was also shed into the medium (Fig. 4B, lanes 1 and 2).
Thus, at steady state ( Fig. 1) or after prolonged radiolabeling, two significant differences were observed in PrP 102L : 1) the C-terminal 18-kDa fragment is significantly decreased, and 2) the C-terminal 20-kDa fragment is increased in PrP 102L cells.
From the pulse-chase experiments, it is clear that PrP 102L was transported to the plasma membrane normally. The marked decrease in 18-kDa may therefore arise because of inefficient endocytosis or recycling of PrP 102L at the plasma membrane. In addition, it is clear from the above data that the over-representation of 20 kDa in PrP 102L is not the result of up-regulation of Ctm PrP, or of an increase in the proteolytic processing of Ctm PrP. Instead, the 20-kDa may accumulate at the plasma membrane because of decreased turnover, probably as a result of inefficient internalization and lysosomal degradation. These possibilities were investigated below.
Endocytosed PrP 102L Is Targeted to Lysosomes Instead of Recycling Back to the Plasma Membrane-The kinetics of PrP C and PrP 102L endocytosis and recycling were examined by immunofluorescence analysis. PrP C and PrP 102L cells were incubated with anti-PrP antibody 3F4 in complete culture medium for 30 min on ice, or for 10 and 60 min each at 37°C in a CO 2 incubator. At the end of each time point, cells were washed with PBS, fixed, and immunostained with anti-mouse RITC to visualize PrP-antibody complexes on the plasma membrane. Excess antibody on the cell surface was quenched with 0.25 mg/ml unlabeled anti-mouse antibody, and the cells were washed and permeabilized with 0.1% Triton X-100. Subsequent immunostaining with FITC-conjugated anti-mouse antibody revealed intracellular PrP-antibody complexes. Thus, with this procedure, PrP molecules on the cell surface stained red, whereas the PrP-antibody complexes internalized during the incubation period stained green.
Following 30 min of incubation with 3F4 on ice, almost all of the PrP C and PrP 102L were detected on the cell surface (Fig. 5A,  red, panels 1 and 2). On the other hand, if the cells were incubated at 37°C for 10 min, most of the PrP C was detected in intracellular vesicles, probably recycling endosomes (Fig. 5A,  panel 3). In contrast, the intracellular vesicles in PrP 102L were concentrated in a perinuclear location rather than close to the plasma membrane (Fig. 5A, panel 4), a difference that became more apparent after 60 min of incubation with 3F4 (Fig. 5A,  panel 5 versus panel 6). Fig. 5B shows confocal images of the same experiment at a higher magnification to emphasize the difference between intracellular localization of endocytic vesicles loaded with PrP C (Fig. 5B, panels 1, 3, and 5) or PrP 102L (Fig. 5B, panels 2, 4, and 6). In addition, there are significant differences in the fluorescence intensity of PrP C and PrP 102L at the cell surface. After 10 min of incubation at 37°C, the surface expression of PrP C and PrP 102L is similar (Fig. 5, A and B,  panels 3 and 4). However, after 60 min, the surface expression of PrP C and PrP 102L was significantly different. Immunoreaction on the surface of PrP C cells showed red and green costaining, indicating that some of the internalized PrP-antibody complexes recycled back to the plasma membrane. In PrP 102L cells, only the red stain was prominent, suggesting that a fair amount of PrP 102L did not get internalized even after 1 h of incubation at 37°C, and that internalized PrP 102L -antibody complexes that should have stained green did not resurface back to the plasma membrane (Fig. 5, A and B, panels 5 and 6).
To check whether the reduced surface expression of PrP 102L is caused by targeting of internalized molecules to the lysosomes, PrP C and PrP 102L cells incubated with 3F4 for 60 min as above were processed for co-immunostaining with cathepsin-D, a marker for late endosomes and lysosomes, or LysoTracker, a marker for lysosomes. As opposed to PrP C , a significant amount of co-staining was observed in PrP 102L samples (data not shown).
Together, the above data indicate that, as opposed to PrP C , PrP 102L is probably degraded by the lysosomes following endocytosis and not recycled back to the plasma membrane.  1 and 2). The 20-kDa fragment is more prominent in the PI-PLCcleaved sample and is over-represented in PrP 102L as compared with PrP C (lanes 3 and 4). Deglycosylation with PNGase-F accentuates the difference in the amount of 20-kDa between PrP C and PrP 102L samples (lanes 5 and 6). An additional band of 14 kDa is detected in the PI-PLC-cleaved sample of PrP 102L (lane 4). Immunoprecipitation with 8H4 shows similar PrP C and PrP 102L glycoforms as observed with 3F4 (lanes 7 and 8). An additional 18-kDa fragment is detected in PrP C lysates (lane 7), majority of which is in the PI-PLC-cleaved sample (lane 9). The 18-kDa fragment is under-represented, and the 20-kDa fragment is over-represented, in PrP 102L lysates (lanes 8 and 10). Deglycosylation of PI-PLC cleaved samples accentuates the difference in 18and 20-kDa fragments in PrP C and PrP 102L (lanes 11 and 12). B, medium collected from overnight labeling of PrP C and PrP 102L cells was immunoprecipitated with 3F4 and analyzed. A small amount of the full-length PrP C and PrP 102L forms (lanes 1 and 2) and a 14-kDa fragment are detected in PrP 102L samples (lane 2).

DISCUSSION
The perplexing lack of correlation among neurodegeneration, transmissibility, and PrP Sc accumulation in GSS102L has been difficult to reconcile with the commonly held belief that PrP Sc is the principal pathogenic agent in all prion disorders. In this report, we provide a detailed analysis of the metabolic consequences of PrP 102L mutation in transfected neuroblastoma cells. We show that the processing and turnover of PrP 102L is altered, resulting in decreased expression of the normal 18-kDa fragment, and increased accumulation of the 20-kDa Ctm PrP fragment on the surface of these cells. This change in phenotype may render PrP 102L cells more susceptible to exogenous PrP Sc infection and toxicity by two inter-related mechanisms. First, the 20-kDa fragment may be more conducive to a change in transformation to PrP Sc because it includes an amyloidogenic region of PrP that is disrupted in the 18-kDa fragment. Second, the 20-kDa may amplify the neurotoxic signal of exogenous PrP Sc as observed in the case of Ctm PrP (22), without serving as a substrate for PrP Sc . A similar change in the cellular phenotype of PrP101L transgenic mice may partially explain the confounding biological effects of this mutation in vivo.
In our cell model of PrP 102L , the surface representation of 20-kDa Ctm PrP fragment is 4-fold more, and the 18-kDa 5-fold less, as compared with PrP C cells. We believe that this difference arises from aberrant recycling of PrP 102L from the cell surface. Normally, cell surface PrP C undergoes constitutive recycling between the plasma membrane and endocytic compartments. During this process, PrP C is cleaved at residues 111/112, and a C-terminal fragment of 18 kDa is transported back to the cell surface (15). Thus, under steady state conditions, 40 -50% of PrP C on the plasma membrane comprises the 18-kDa fragment rather than full-length PrP C . This observation has important clinical implications, because the 18-kDa is disrupted at residues 111/112 and therefore lacks critical residues, especially 90 -112, involved in the refolding of PrP C to PrP Sc (27)(28)(29)(30). The significance of these residues is highlighted by the fact that, in PrP-null mice that are resistant to scrapie infection, transgenes encoding PrP with deletions up to residue 93, but not 106, restore susceptibility of the animals to PrP Sc infection (25). Because the 20-kDa fragment, but not the 18-kDa fragment, includes most of this region, a change in the ratio of 20-to 18-kDa fragment on the surface of PrP 102Lexpressing cells could have profound implications for PrP Sc replication and neurotoxicity in vivo.
In our cell model, PrP 102L is neither aggregated nor PKresistant. Although this observation is consistent with reports on PrP101L-overexpressing transgenic mice that develop neurologic disease in the absence of detectable PrP Sc (11,13,14), the mechanism by which mutant PrP mediates neurodegeneration is unclear. Adding to this apparent paradox are studies on transgenic mice expressing only a single allele of PrP101L, and the PG14 mice with an expanded octapeptide repeat that fail to develop spontaneous neurodegenerative disease, but exhibit increased vulnerability to exogenous PrP Sc infection (13,14,26). When considered in context with our data, it is reasonable to assume that enhanced susceptibility of PrP101L mice to PrP Sc infection may be the result of increased expression of the potentially amyloidogenic 20-kDa fragment on the cell surface, providing an optimal substrate for the incoming PrP Sc . Once a nidus of PrP Sc is initiated on the cell surface, albeit comprising only the 20-kDa fragment, subsequent conversion of full-length PrP and additional 20-kDa fragments would occur exponentially, as reported in our earlier study (16). Such a scenario would account for the increased susceptibility of PrP101L transgenic mice to exogenous PrP Sc infection, as reported by Barron et al. (14). Alternately, the 20-kDa fragment may am- FIG. 5. Endocytosed PrP 102L is targeted to lysosomes instead of recycling back to the plasma membrane. PrP C and PrP 102L cells were incubated with anti-PrP antibody 3F4 in complete culture medium for 30 min on ice or for 10 and 60 min at 37°C under normal culture conditions. At the end of each time point, the cells were washed with PBS, fixed, and immunostained with anti-mouse RITC to visualize PrP-antibody complexes on the plasma membrane. Excess antibody on the cell surface was quenched with 0.25 mg/ml unlabeled anti-mouse antibody, and the cells were washed and permeabilized with 0.1% Triton X-100. Subsequent immunostaining with FITC-conjugated antimouse antibody revealed intracellular PrP-antibody complexes. A, following 30 min of incubation with 3F4 on ice, almost all of the PrP C and PrP 102L are detected on the cell surface (panels 1 and 2). Incubation at 37°C for 10 min shows most of the PrP C in intracellular vesicles close to the plasma membrane (panel 3). In contrast, intracellular vesicles in PrP 102L are concentrated in a perinuclear location (panel 4). This difference is more apparent after the 60-min time point (panel 5 versus panel 6). B, confocal images of the above experiment at a higher magnification to emphasize the difference in intracellular localization of PrP C (panels 1, 3, and 5) and PrP 102L (panels 2, 4, and 6). After 60 min of incubation with 3F4 the surface expression of PrP C is significantly more than PrP 102L (panels 5 and 6). plify the neurotoxic signal initiated by exogenous PrP Sc without itself undergoing a transformational change, causing neurotoxicity by a Ctm PrP-mediated phenomenon. The latter scenario is supported by studies where an inverse correlation between PrP Sc accumulation and Ctm PrP has been described in mice challenged with PrP Sc , implicating Ctm PrP as the principal modulator of neurotoxicity (22).
Based on the results reported here and in a previous study, the 20-kDa fragment is generated in the ER probably from Ctm PrP (16), whereas the 18-kDa fragment is produced during recycling from the cell surface (15). Although PK treatment of microsomes prepared from PrP C and PrP 102L cells resulted in an increase in the amount of 20-kDa fragment, this result only confirms the origin of 20-kDa fragment from Ctm PrP. The amount of 20-kDa fragment generated in the two cell lines is not altered significantly, and thus the increased representation of 20-kDa fragment in PrP 102L cells cannot be attributed to an up-regulation in the synthesis, or increased proteolytic cleavage of Ctm PrP. Instead, we believe that the accumulation of 20-kDa fragment on the surface of PrP 102L cells is caused by decreased turnover. Our results on the kinetics of endocytosis and recycling of PrP C and PrP 102L show clearly that, whereas PrP C remains in peripheral endosomes close to the plasma membrane and is recycled back to the cell surface, most of PrP 102L accumulates in perinuclear structures, probably lysosomes, and is degraded. These results explain the decreased expression of 18-kDa fragment on the surface of PrP 102L cells, because 18-kDa fragment must recycle back to the plasma membrane after cleavage in an endocytic compartment. The mechanism leading to over-representation of 20-kDa fragment on the surface of PrP 102L cells is not entirely clear from our data. Because significant amounts of this fragment are observed only at steady state or after prolonged radiolabeling, it appears to accumulate over time, probably because of slow turnover. Because the 20-kDa fragment in PrP 102L cells migrates faster on SDS-PAGE than the corresponding fragment from PrP C , it probably includes the PrP 102L mutation. The faster migration in the 20-kDa and not the 18-kDa fragment of PrP 102L is probably a result of the amino acid change to leucine, allowing better binding to SDS than proline, which is a helix breaker. It is possible that the 20-kDa fragment with the PrP 102L mutation is endocytosed less efficiently because of the mutation, which may cause a change in its conformation. A similar change in secondary structure may account for the lysosomal delivery of PrP 102L after endocytosis rather than recycling back to the plasma membrane. Such an alteration in the targeting and turnover of full-length PrP 102L and its 20-kDa fragment would reverse the ratio of 20-to 18-kDa fragment on the surface of these cells, thus altering their phenotype in a significant way. However, both full-length PrP 102L and its 20-kDa fragment can be released from the cell surface by PI-PLC, excluding the possibility that these are grossly misfolded or aggregated on the cell surface.
Although the 20-kDa fragment arises from Ctm PrP, which is spared by proteasomal inhibition (20), the origin of 14-kDa fragment is unclear. Antibody mapping studies show that it is an N-terminal fragment and is secreted into the medium under normal conditions, and more so when proteasomal function is inhibited. 2 It is plausible that the 14-kDa arises from Ntm PrP, another transmembrane form of PrP with the C-terminal domain in the cytosol, and the N-terminal 113 amino acids in the ER. However, unlike the 20-kDa fragment, PK treatment of microsomes does not result in an increase in the amount of 14-kDa fragment, arguing against its origin from Ntm PrP. Further investigations on the identity of this fragment are under progress.
In conclusion, this report highlights important differences in the metabolism of PrP 102L as compared with PrP C , implicating the 20-kDa metabolic product of Ctm PrP in the pathogenesis of GSS102L. Although the underlying mechanism of the pathogenic process is presently unclear, this report provides the basis for future investigations to clarify the complex biological manifestations of this mutation in vivo.